Sequence Data Analysis and Quality Filtering
A total of 1,262,784 paired-end reads were generated from five cecal samples of study groups. After quality filtration, 1,083,070 reads remained with average read number per chicken 96,823 (± 34,814 s.d.) having median read length of 458 (± 9 s.d.) bases in all samples. Altogether, 1,195 distinct OTUs at 97% similarity index threshold were obtained from all samples, out of which, 895 remained for downstream analyses.
Dietary Variation in the Cecal Microbiota
Rarefaction analysis of OTUs from cecal samples indicated good sequencing coverage (~ 90%) in all samples (Figure 1). The cecal microbial diversity among study groups was measured through alpha and beta diversity. Alpha diversity was estimated by using Chao1, ACE, Shannon, Simpson, and number of observed species (Table 2). Alpha rarefaction curves of observed species, chao1, ACE, shannon and simpson against number of sequences in study groups are shown in (Online Resource 1). The ACE and Shannon indexes described higher species richness (p<0.05) in COM, AB, and ORG groups. Simpson index illustrated indicated species evenness (p<0.05) in phytogenic feed additive (PHY) group. Furthermore, the higher bacterial diversity was achieved in COM and AB groups. Beta diversity analysis were executed through weighted/unweighted UniFrac and Bray-Curtis distances by using PERMANOVA and UPGMA tests to generate heat map and cluster tree. Bray-Curtis and weighted UniFrac distances are quantitative beta diversity traits showing significant differences (p<0.001) between CON and feed additive (AB, ORG, PHY and COM) groups. Cluster analysis revealed clear distinction related to abundance of similar species in CON and PHY groups (Figure 2a). Unweighted UniFrac diversity distance illustrated pronounced microbial community presence in PHY group (p<0.001) and cluster analysis also placed PHY group in a separate node (Figure 2b). Similar results with Bray-Curtis diversity index was achieved placing PHY in separate branch with CON (Figure 2c). The diversity differences were further exemplified by principal component analysis (PCA) which demonstrated clear demarcation among bacterial assemblages of all groups along principal component axis 1 of PCA plot. The total variance of 83.96% was apparent in cecum samples along the two axes of PCA (Figure 2d). Given that dietary variation contributed to the shifts in microbial communities in chicken cecum, we further evaluated the shared/unique OTUs in cecum samples (Figure 2e). In total, 31 OTUs were shared across all study groups, accounting for 3.5% of total OTUs, which represents “core” microbiota in chicken cecum. The analysis of shared microbial communities between each study groups were also analyzed as shown in figure (Online Resource 2).
Hierarchical Clustering of Microbiome in Chicken Cecum Across Dietary Treatments
The bacterial community distribution pattern under various taxonomic classification levels encompassing phylum, class, order, family, genus, and species were compared against RDP classifier database using PCA. The total quality filtered reads (1,083,070) from cecal samples were assigned to 1,195 OTUs, which further distributed into ten phyla and one unclassified phylum (Figure 3). The dominant phylum in chicken cecum was Firmicutes in all groups with higher abundance in PHY (99%) and CON (64%) groups. The second most abundant phylum was Bacteroides followed by Proteobacteria, Deferribacteres, Synergistetes and Verrucomicrobia with relative abundance of 24.6%, 9.8%, 4%, 3.8% and 2% respectively. Bacilli (70-90%) was the abundant class in PHY and CON groups from phylum Firmicutes, with the members of order Bacillales being most prominent. Interestingly, a diverse bacterial community distribution at class level was noticed in COM, AB, and ORG groups. Clostridia was the most abundant class in all study groups with relative abundance of 56%, 50% and 35% in AB, ORG and COM groups respectively. Class Bacteroidia was the third most abundant class in chicken cecum (Figure 4). Order Bacteroidales and Clostridiales were the abundant orders in COM, AB, and ORG groups. The presence of orders Acidaminococcales and Burkholderiales were also noticed in ORG (Figure 5). Notably, a large proportion of unclassified OTUs at order level were seen in COM and ORG groups. The taxonomic composition at family level showed highly diversified bacterial community in AB, ORG and COM groups (Online Resource 3). Campylobacteraceae and Bacteroidaceae (20-30%) were prominent families in AB. At genus level, PHY group was dominated by Bacillus (99.2%) from family Bacillaceae (Online Resource 4). The abundant bacterial genera in all samples except PHY were Helicobacter, Blautia, Ruminiclostridium, Campylobacter, Desulfovibrio and Bacteroides. The only species found in PHY group was Bacillus_safensis_FO-36b (Figure 6). The dominant species in CON and AB was Bacteroides_barnesiae (20-30%). ORG group showed high abundance of unclassified species (58%) along with Helicobacter_pullorum_NCTC_12824 (18%), Ruminococcus_torques_ATCC_27756 (7%) and Mucispirillum_schaedleri (9%).
Abundance Differences of Microbiota Among Study Groups
To further investigate the differences in relative abundance of microbiota and their association with the study groups, heat maps were drawn. The heat map showed presence of all bacteria at phylum level (p<0.05) in groups (Figure 7a) whereas species showing low abundance (less than 0.5) were later classified as ‘others’ in subsequent ranks. Phytogenic feed additives showed highest value of 1 for phylum Firmicutes and formed distinct branch depicting presence of discreet bacterial phyla composition. Based on the occurrence of microbiota at class level, COM and ORG were placed in a similar branch in clustering analysis showing relative abundance of similar bacteria (Figure 7b). At order level, AB and ORG groups shared phenetic relatedness (Figure 7c). Similarly, presence of order Verrucomicrobiales was only noticed in control and combination groups. The presence of family Akkermansiaceae was noticed in COM group only with high abundance (Figure 7d). The pattern of bacterial families present in AB and ORG group were similar in relative abundance. In COM, presence of genus Akkermansia (family Akkermansiaceae) was observed (Figure 7e) with the presence of species Akkermansia_muciniphila_ATCC_BAA-835. Surprisingly, AB and COM groups shared similar branch in heat map at species level depicting species richness in both groups (Figure 7f) thus, validating the synergistic effect of natural feed additives in maintaining bacterial diversity in chicken cecum.
Phylogenetic Analysis of Dominant Bacterial Genera in Chicken Cecum
Besides species composition and abundance analysis, a cladogram was constructed from the tags of abundant bacterial genera to elucidate the species relatedness present in chicken cecum (Figure 8). Majority of the genera belonged to phylum Firmicutes (56%) which was bifurcated into two branches. The first branch having genus Ruminiclostridium showed evolutionary relatedness with genera from Proteobacteria (15%) whereas the second branch (Lactobacillus, Enterococcus) formed a distinct clade. Phylum Bacteroides, second most abundant phylum (17%), formed separate clade depicting phenetic relatedness with phylum Lentisphaerae. Phylum Deferribacteres and Synergistetes were closely related with each other while, Elusimicrobia formed a discreet branch in cladogram.